Most of us here are familiar with the Humean view that only desires set ends in action (or play this role in motivating and rationalizing action). Disambiguating, precisifying and focusing on the rationalizing element, we get: only intrinsic desires set the ultimate ends for rationalizing intention (and hence action).
By way of contrast, we might note that instrumental rationality, broadly speaking, accounts for the formation of instrumental desires (in the broad sense that includes realizer desires) and the formation of intentions to act (either presently or in the future). We might further distinguish the formation of singular intentions from the formation of whole structures of intentions (including conditional intentions) that are plans (and of course plans, or parts of plans that are themselves mini-plans, are often conditional as well). All of this, it seems is the result of “instrumental reason” – our various capacities for being instrumentally rational – acting on intrinsic desires to produce the further attitudes necessary to generate appropriate motivation and intention (including planning) given one’s relevant instrumental information. As A&S point out, some of these processes, and the most basic of them, are performed automatically and do not require much or any reflective self-awareness in the process (and hence are not acts of deliberation).
Though I often find this picture compelling (and not due to simply defining intrinsic desires as the things that set the ends that instrumental reason operates on), I wonder about the potential role of various evaluative and normative judgments that might be: (a) in the business of the same kind of end-setting that intrinsic desires are and (b) themselves the product of instrumental rationality.
In the latter functional category, consider first-personal judgments about what one ought to do, where the ‘ought’ here is a summation of the relationship of the action to various antecedent ends (whether these are set by intrinsic desires or by something else). It seems to me that such beliefs (or belief-like attitudes) function much like intentions. More precisely, they seem to constitute a cognitive channel for generating intentions, plans, etc. And their production quite plausibly involves instrumental rationality operating on antecedent ends – perhaps set or constituted by antecedent intrinsic desire, though it is perhaps more intuitive that they are set, in part, via often via relevant sorts of evaluative or normative belief.
And this brings us to the former functional category, the category that concerns ultimate-end-setting). Consider two types of evaluative or normative judgments: (A) the first-personal judgment that one ought to do something simply because it is inherently very, very much worth doing (simply because one ought to do it, as we sometimes say) and (B) the judgment that something is inherently desirable. It seems that, at least sometimes or in some contexts, such judgments set ends in something like the way that intrinsic desires do. And they don’t seem to do this, at least not typically, by generating accompanying intrinsic desires. This is something that we have to explain (or at least explain away). Simply asserting that only intrinsic desires play this role will not do.
We can say a bit more. Focus on judgments of inherent desirability. Even if – as I think is plausible – desirability judgments evaluate (possible, consequent) intrinsic desires ultimately on the basis of distinct (actual, antecedent) intrinsic desires, desirability-judgments seem to both (1) play a kind of present-ultimate-end-setting role and (2) guide, insofar as rationality can, the process of forming additional intrinsic desires and in some sense improving our overall profile of intrinsic desires (relative to coherence and relative to our basic personal and social functional architecture as humans). The first thing here indicates an alternative motivational material for a broadly instrumental rational process to work on. The second thing here indicates an executive, regulative function with respect to (possible, consequent) intrinsic desire. Even if this executive, intrinsic-desire-shaping function ultimately depends utterly on antecedent intrinsic desires, this role would seem to be important (obviously, this provides a different explanation for certain scenarios of getting out of bed to go to work even though you don’t really want to than A&S can avail themselves of).
So I’m inclined to buck Hume with regard to Reason (our cognitive rational processes and its products) being able to set the ultimate ends of present practical rationalization. Judgments of inherent desirability (and perhaps some others) can do this. And, in doing this, they may improve planning and action by allowing action to be guided not only by the intrinsic desires that we do have, but also (and preferentially) by the intrinsic desires that we ought to have (via evaluative or normative judgments with this content). In addition, the output of broadly instrumental rational response can be (instrumental) evaluative or normative judgments appropriate to intention-formation and planning. In this way, Reason can, and sometimes does well to, override what would be the inputs and what would be the outputs of practical reasoning without evaluative or normative judgments (or without their playing this role). Of course, lots of details to work out, here. And maybe I’m wrong. However, if I am right, I remain a Humean in this sense: the ultimate ultimate ends, the ends that along with relevant functional and circumstantially causal information allow us to determine that P is an appropriate object of intrinsic desire, are set by (actual, antecedent) intrinsic desire.
A functional view of desire like A&N’s helps us explain how things like desirability judgments might work. Part of their contents – the that-P part of that-P being inherently desirable – are functionally connected to motivation (but not emotional reaction or attention/cognition). Though there might be good explanatory reason to say that, in this case, one intrinsically desires that-P this seems odd considering that the content of the judgment allows that one does not presently intrinsically desire that-P. (This is a purely conceptual reason to be skeptical about making the that-P part of the that-P-being-inherently-desirable out as an intrinsic desire. But I suspect – and hope, for the sake of the adequacy of this part of our conceptual scheme to the world – that distinctions drive by good explanation would not countenance lumping the motivational element of the belief that P is inherently desirable into the category of desires, the intrinsic desire that-P.)
I see no reason why desires could not be contents that are constituted as rewards by the reward learning system of the brain (or some specific brain system of this general type – I suspect that A&N are specifying a type of learning mechanism that is too general in their version of this sort of account – despite being well in the right ballpark for “reductively” explaining what desires are in terms of brain process and function). Maybe that-P getting constituted as a reward in the right kind of reward-learning system constitutes intrinsically desire that P – but when we judge that that-P is inherently desirable that that-P part of the content does not get thus constituted. Broadly speaking (abstracting from the neuroscience that I am ignorant of), there would have to be a relevant difference in function (or constitutive functional profile) to support this difference. It is easy to think of plausible candidate distinct functional profiles a priori, but it would be better to do so in a scientifically-informed not just conceptually-informed way. Maybe or maybe not to do we have the right science done to complete this task. But we can hammer away at both the conceptual and empirical sides of the issue (each informed by the other or we’ll likely go off the rails) until we get it right.
In any case, I suspect that A&S’s radically Humean psychology of acting for reasons (or rationalizing action) will not do. In effect, they have achieved via empirical hypothesis what is sometimes achieved via definitional fiat (in declaring that desiring that-P is simply that-P having certain motivating or affect-inducing tendencies, etc.). The result, however, is the same: the judgment that it is best for one to get up and go to work rather than stay in bed becomes (in part) a special, “cold” sort of desire (connected to motivation, not so much to affect). On the sort of view that I am suggesting, “cold” desires are simply desires (contents with the right functional profile, at least aiming at both motivation and appropriate affect) that are blocked from doing their full job – whereas, in contrast, the that-P part of the inherently-desirable-that-P judgment simply does not have the same job (same function, same constitutive functional profile).